Bauer

A proof-of-concept Go application that extracts document content, suggestions (proposed edits), and comments from Google Docs using the Google Docs API and Google Drive API.

Installation

Snap

sudo snap install bauer

Homebrew

First time installation

brew install britneywwc/bauer/bauer

Upgrade to a newer version or later

brew update
brew upgrade bauer

N.B. You need to install Copilot CLI which is used by Bauer.

Configuration

1. Install Copilot CLI
2. Create credentials.json file and copy the structure from the default file
3. Get credentials from Google Cloud service or Bitwarden (internally)
4. Fill up credentials.json with Google Cloud credentials (see Generating Google Cloud credentials).
5. Share copy document with service account

Usage

1. Install bauer using the instructions above
2. Check that copilot and bauer are installed
3. Get document ID from Google Document & share the document with the service account
4. Run Bauer

bauer --doc-id <your-document-id> --credentials ./credentials.json

6. Optional parameters

Flag	Type	Default	Description
--chunk-size	int	1	Total number of chunks to create (default: 1, or 5 if --page-refresh is set)
--dry-run	bool	false	Run extraction and planning only; skip Copilot execution and PR creation
--output-dir	string	bauer-output	Output directory for generated files
--model	string	gpt-5-mini-high	Copilot model to use for code generation
--page-refresh	bool	false	Whether this is a page refresh, or the default copy update

Examples

Basic run

bauer --doc-id <your-document-id> --credentials ./credentials.json

Dry run (test without executing changes)

bauer --doc-id <your-document-id> \
        --credentials ./credentials.json \
        --dry-run

Custom chunk size and output directory

bauer --doc-id <your-document-id> \
        --credentials ./credentials.json \
        --chunk-size 5 \
        --output-dir ./results

Specify model

bauer --doc-id <your-document-id> \
        --credentials ./credentials.json \
        --model "claude-sonnet-4.5"

Page refresh

bauer --doc-id <your-document-id> \
        --credentials ./credentials.json \
        --page-refresh

Local development

Prerequisites

1. Install go
2. Install task
3. Install Copilot CLI

Steps

1. Modify the Taskfile with your document ID and credentials path for convenience
2. Run the project with task

task run

Documentation

For more information refer to ARCHITECTURE.md

Future improvements

Short term

• Automatically open PR with changes applied to the document using Google Docs API
• Improve prompt templates for better results (this requires a lot of trial and error)

for code improvements, you can also refer to our todo list

Long term

On the long term, BAUer should evolve into a full-fledged API service, with the following features:

• Automatic Jira ticket hooks to trigger workflows
• Unified service account with domain wide delegation
• Calling LLMs - with varying implementation complexity - via: - calling LLM APIs directly - spinning up ephemeral Copilot CLI instances - self-hosted LLMs (can use open source models such as Llama, openAI OSS, deepseek, etc)
• Automatic PR creations and reviewer assignments